Structure Extraction in Task-Oriented Dialogues with Slot Clustering

TL;DR

This paper presents a method for extracting dialogue structure in task-oriented conversations by clustering slot tokens with a pre-trained model, enabling better understanding and improved response generation without manual state annotation.

Contribution

The authors introduce a novel approach combining slot clustering and tracking to automatically derive dialogue structures, enhancing efficiency and performance in dialogue systems.

Findings

Outperforms unsupervised baseline models in structure extraction

Data augmentation using extracted structures improves response generation

Effective approximation of dialogue ontology without manual annotation

Abstract

Extracting structure information from dialogue data can help us better understand user and system behaviors. In task-oriented dialogues, dialogue structure has often been considered as transition graphs among dialogue states. However, annotating dialogue states manually is expensive and time-consuming. In this paper, we propose a simple yet effective approach for structure extraction in task-oriented dialogues. We first detect and cluster possible slot tokens with a pre-trained model to approximate dialogue ontology for a target domain. Then we track the status of each identified token group and derive a state transition structure. Empirical results show that our approach outperforms unsupervised baseline models by far in dialogue structure extraction. In addition, we show that data augmentation based on extracted structures enriches the surface formats of training data and can achieve a significant performance boost in dialogue response generation.